Report Summary: Expanding Supercomputing Availability Across America Through Federal Government Efforts

The Center for Data Innovation recently hosted a panel discussion on December 10, exploring ways to enhance access to High-Performance Computing (HPC) resources for Artificial Intelligence (AI) researchers in the United States. Hodan Omaar presented the Center's report titled "How the United States Can Increase Access to Supercomputing."

The report identifies three main problems limiting AI researchers' access to HPC resources. Firstly, long-term initiatives are needed to address the leaky pipeline of underrepresented groups with computer science backgrounds entering the HPC workforce. Secondly, access to HPC is concentrated in states with leading academic institutions, making it difficult for researchers in states with fewer leading universities to pursue their research goals. Thirdly, currently, HPC resources invested by the National Science Foundation (NSF) and the Department of Energy (DOE) can only support a third of the researchers who need access to them.

To address these issues, panelists emphasized the need for the United States to take significant steps today. They suggested expanding and supporting programs like NSF’s National AI Research Resource (NAIRR) Pilot and Advanced Cyberinfrastructure Coordination Ecosystem (ACCESS), as well as DOE’s INCITE program. These initiatives would increase AI researchers’ access to high-performance computing by federating government supercomputing clusters and partnering with private firms for additional cloud computing resources.

Andrew Jones, planning future HPC and AI capabilities at Microsoft, emphasized the importance of funding being matched with strategic planning and operational processes, coordinated between public and private stakeholders at both the local and national level. The Coalition for Academic Scientific Computation, which brings universities and computing centers together to share HPC knowledge and tools and advocate for the needs of the HPC community, was also highlighted as a crucial player in this endeavour.

Panelists agreed that Congress should significantly increase funding for HPC to both NSF and DOE. Lack of federal funding in HPC prevents these organisations from investing in building an adequate supply of HPC resources for researchers. Without a clear strategy for acquiring state-of-the-art computing infrastructure, software, and experts, disparate initiatives may impede researchers from being more productive and innovating at higher rates than their competitors.

The gender and diversity imbalance in the HPC community is hindering the entry of qualified candidates from underrepresented groups into the workforce. Women and minorities are underrepresented in the HPC workforce, which in turn limits their ability to become the next generation of AI researchers. Short-term initiatives are needed to make underrepresented groups aware of available resources and to better represent Minority-Serving Institutions (MSIs) in larger organisations.

Nvidia's director of global business development, higher education, and research suggested that the funding figure will need to grow over time as AI applications expand into new disciplines. A notable example of university-industry-government partnerships is Nvidia's partnership with the University of Florida to launch an AI supercomputer.

The Center's report also suggests exploring cloud computing options to diversify HPC resources for AI researchers. Providing access to cloud services for AI researchers across the country can provide more researchers with the tools they need. The report underscores the need for the United States to ensure its AI researchers have the freedom and ability to choose the best HPC resources.

In conclusion, increasing access to HPC resources for AI researchers is crucial for the United States to remain competitive in the global AI landscape. By addressing the leaky pipeline, expanding funding, and exploring cloud computing options, the United States can ensure its AI researchers have the tools they need to drive innovation and discovery.

Report Summary: Expanding Supercomputing Availability Across America Through Federal Government Efforts